I just received a copy of Robert Plotkin’s new book: The Genie in the Machine: How Computer-Automated Inventing is Revolutionizing Law and Business. The book is just out from Stanford Law Books. It is a great read for a good price ($20.00 from Amazon).
Plotkin’s new book begins with the reality: many products and processes used today were created through automated software. In Plotkin’s book, the automated software is akin to a genie granting the wishes demanded by software developers. The trend toward automated inventing will likely continue to grow as both computing power and the level of problem-complication increase.
Plotkin is a patent attorney and has been a leader of software law for the past decade. Most of the book is spent considering how patent law interacts (and should interact) with computer created inventions. Plotkin believes that automated computer-generated inventions should be patentable. However, he would make some changes to ensure that the computer inventions are not overbroad.
“We need . . . to reinterpret [the patent] rules in light of artificial invention technology to ensure that they continue to achieve their original goals, just as a butcher recalibrates a scale after it has been in long use . . . In particular, we need to strictly apply patent law’s utility requirement and its distinction between abstract ideas and practical applications [to] protect against the worst kind of abuses”
If a computer creates an invention, then it should have the right to exclude others from making, using, and selling that invention. Then all the money made from that invention can be funneled into a black hole useable only at the discretion of the computer. One step closer to humanity’s enslavement by computers.
Automated invention and TRIZ and the implications of both on the nonobviouness standard have been addressed with in my article entitled “Reinventing Invention” in Volume 91, Issue 2 of the Journal of the Patent and Trademark Office Society (Feb 2009). There are some significant implications here – and we will evolve to address them. The only question is how quickly.
“peanut gallery” said:
“I think that it would be more accurate to say the the computer and/or operator discovered, rather than invented. However, I’m not sure what legal significance whould be given to that distinction, since patents are supposed to be for both discoveries and inventions. (as long as the discoveries are not natural phenomena)”
An astute observation. While it is true that he US constitution effectively says patents are for discoveries, in other countries there is a more clear distinction between discoveries and inventions in terms of patentability. In the UK for instance, mere “discoveries” are not patentable; one must translate the discovery into something with utility, and what not. Of course, the US also has utility provisions too, so that “mere” discoveries are in fact not patentable.
But differences remain, not least because of the influence of TRP, etc. I could well imagine how my new chemistry reaction example would be called a discovery on both sides of the pond, and that as such it might might not be patentable in the UK (absent a little work getting around the utility issues), while being patentable in the US – it won’t be a “natural” phenomenon, since it occurs only in the lab (once it is taken ex silicio), not in nature.
IIRC “Element 95” was a successful claim in the US, indeed, it survived litigation in In re Seaborg (CCPA) in 1964 (on an inherency issue). I don’t know if it would be successful now in the US, given current utility standards, but it I don’t think it would be be in the UK, since it looks like a good case of a “discovery” that is not an invention as opposed to one that is.
A computer-predicted reaction could be another such example?
Cheers, Luke
“inviting body punches” asked:
“an invention having, in the eyes of the law, no inventors”–How did you reach this conclusion? A patent application cannot be filed without identification of the inventor(s), therefore a patentable invention MUST have an inventor, in the eyes of the law.
Because I said “invention”, not “patentable invention” – and certainly not “patentable as defined by one particular statute currently in force in one particular country”.
“The programmer most certainly DID know what the program would do in the general case”.
But I was talking about a specific case. And we are not talking about what the program will do (churn numbers) but what the program will produce (in this case, an invention or a discovery as you may prefer; see next post).
“The program is entirely deterministic–even if it incorporates something like a PRNG, the selection of that number is ordered, and any action to be taken upon, or as a result of, such number, is pre-determined. Even if conditionals such as if-then-else are used, the selection is again ordered, and each possible future course is pre-determined”.
Irrelevent. (If you say randomness is doesn’t void determinism because you can hypothetically produce every conceivable state, given infinite time, then so is the entire universe, and thus there is no such thing as an invention)
“From the point of view of the programmer, the particular result was not unexpected, only unknown, and only because the programmer was unaware of the initial conditions specified”.
Can’t agree with that. The initial conditions specified were “known” – “selected” you would say. The result was not only unknown, not only unexpected, but also unforseeable.
Cheers, Luke
“The computer did invent, just because it brute force calculated the invention does not make it less of an invention.”
I think that it would be more accurate to say the the computer and/or operator discovered, rather than invented. However, I’m not sure what legal significance whould be given to that distinction, since patents are supposed to be for both discoveries and inventions. (as long as the discoveries are not natural phenomena)
Luke–
Fun discussion. Various responses to your post:
“an invention having, in the eyes of the law, no inventors”–How did you reach this conclusion? A patent application cannot be filed without identification of the inventor(s), therefore a patentable invention MUST have an inventor, in the eyes of the law.
The programmer most certainly DID know what the program would do in the general case. The program is entirely deterministic–even if it incorporates something like a PRNG, the selection of that number is ordered, and any action to be taken upon, or as a result of, such number, is pre-determined. Even if conditionals such as if-then-else are used, the selection is again ordered, and each possible future course is pre-determined.
From the point of view of the programmer, the particular result was not unexpected, only unknown, and only because the programmer was unaware of the initial conditions specified. The particular result, as forecast by the programmer, would be that result that the program would achieve given a particular set of input data.
There is ALWAYS selection of data, even if you select the entire available set. Firstly, you are choosing not to exclude any data in the set–second, and more importantly, you have chosen which data are within the set.
“There is no guiding will/hand”. That cannot be true–software, hardware, and the interplay between the two do not exist but for the action of human consciousness. They are man-made.
When I implied that a cognition, autonomy, volition, and conception are required to “invent”, I was speaking in legal patent terms. Certainly something “new, useful, and unobvious” can arise from other than this source, but such things are excluded from the ambit of patent protection as naturally-occurring phenomena. Inventions that arise in such a way may be “inventions”, but they are not “patentable inventions”.
I prefer to think of them not as “inventions”, and to reserve that term for the products of human action. They can be thought of as “products” of evolution or natural circumstance. Certainly neither hardware nor software are naturally-occurring.
If the computer program can account for the conditions giving rise to the “invention,” which it must necessarily, then there is a good chance that the “invention”is obvious depending on when the computer program was written. But in any case its not the computer doing the inventing, but the developer creating the conditions that lead to the “invention.”
“As I’ve pointed out a billion times, however, that is only true where the cost of developing the new product is so high relative to the cost of manufacturing and marketing the product that a patent is necessary for anyone to profit from the sale of the product”
Mooney never pointed this out “a billion times,” Mooney waited for others to point this out and then co-opted it. Mooney is making stuff up as he goes along because he doesn’t really know spit.
And I’m being kind.
Luke
Interesting points. My central point, I think, would be that whatever role the computer played, there will always be humans in the loop and those humans are candidates for inventorship, depending on their actual contribution to the process that produced the invention.
Interesting question also about Mandelbrot images, speaking as someone who wrote my own Mandelbrot program many years ago, in GFA Basic on an Atari ST.
“ThisMortalCoil” wrote:
“I guess the over-riding importance of who *is* the inventor is peculiar to US patent law. The real question must surely be: who is entitled to ownership of the rights?”
The last part is (mostly) correct; the first part is not – the problem is identical for all varieties of patent law. I say *mostly* correct – because you are making an assumption that there will be somebody entitled to the ownership of the rights, whereas I would posit that this is not necessarily the case – i.e. you can have an invention without an inventor.
IIRC the US consitution awards patents to *inventors* who make “discoveries” by whih we understand “inventions” – but that clearly doesn’t imply all “discoveries” (read “inventions”) are made by “inventors”.
“Suppose you developed a program that generated “inventions” on the basis of information fed to it by human beings. Is there any reason to question that the devisers of the program together with the individuals who decided on the input ought to be the owners?”
Yes. For one thing, the data need not be fed by a human – programs can gather there own data too. Inventorship should not, IMO, turn on whether the *the very same data* being fed into a computer is being fed manually by a human or being fed automatically by a(nother) computer. The data feeder is just a drone, one way of the other. *If* there is some inventive selection going on in the data feeding I would agree with you. But you are assuming there is – and this is a clearly false assumption in some cases. Your scenario is assuming an “individual” has “decided” on the input. Neither of these assumptions are sound.
“I’m sure that more highly nuanced scenarios could be posited, but I don’t see any great difficulty in applying the idea that whoever enabled the machine-conception of the invention should own the rights”.
Because they may be completely unaware that any invention has been made. They will have sold the program to a 3rd party and the 3rd party will have used it to generate something valuable. Why should the 1st party reap the windfall for all that downstream activity – activity that they don’t even know about? I don’t see any significant difference here between the situation with ISPs owning or not internet content generated by others.
“I also don’t see any particular problem with identifying the “humans-in-the-loop” as being the “authors”, to satisfy the US constitution”.
If we are venturing into copyright here, who owns the copyright to a Mandlebrot picture I make using some software? Me, or the software writer? I assume you would generally say me. But does it make a difference if I didn’t enter any of the input parameters, and use the default settings that came with the program? Is that now the vendor’s picture – or is it no longer an “original” work? Probably not original, so no copyright at all. But now, let’s explore further – does it make a difference if the default “settings” that came with the program where random generated numbers ranging from -1 to 1? I believe it does. Each setting is unique – and thus clearly “original”. The software writer certainly didn’t chose the starting configuration (heck – they just implemented a fairly basic mathematical formula). But neither did I; all I did was press “enter” with zero thought required. I could have had my cat walk over the keyboard until it accidentally hit the enter key.
Can it be that in an original work, there is no “author” at all?
IMO, yes, there can be, just as I believe that currently, you can have an invention without having a legal “inventor”.
And I further believe there could well be sound policy reasons for allowing such a situation to exist – but the implications of “computer-generated” works and inventions have hardly begun to be explored.
Cheers, Luke
“inviting body punches” asked me:
“Luke–
Why could there not be multiple inventors?”
That is not the problem; the problem is what I see as a clear “invention” having, in the eyes of the law, *no* inventors.
“At a minimum, I would suggest that both the programmer and the person who selected the particular data input would be co-inventors.
The programmer knew what the program would do in the general case, the person who selected the particular data input defined the parameters of the particular case.”
No, the programmer did not know what the program would do, that is the whole point. That is why the result was unexpected. If you knew what the result is going to be before doing the experiment, there is little need to do the experiment.
And as for “selecting the data” – who said there was any “selection”? E.g. inputting an *entire* data set means no selection at all, although selection will be required if you delete some items from that set first.
“Neither one acting alone would have achieved the particular result, but both acting together did.”
Which is true of most inventions, whether involving software of not – but the law likes to pretend otherwise.
“The computer did not invent. It has no volition. It cannot conceive. It does not cognate. It is not autonomous. It exists because of, and acts according to the direction of, a conscious mind”.
I can’t see the logic behind that statement. You seem to equate the capacity to invent to conciousness. I would like to see some reasoning behind this equation.
“In a sense, it could be analogous to “the guys in the shop”, or the grad student, who actually executes the will of a researcher in achieving a particular solution”.
But here, it is my position that *there is no researcher*. There is no guiding will/hand.
Cheers, Luke
Regarding incentive to invent, I think patents more likely offer incentive to disclose where trade secret is possible, incentive to bring the product to market where the product is not profitable without monopoly protection, and incentive to reduce the invention to practice where the expense of doing so outweighs all profit obtainable in the presence of free market competition. In the first case, trade secret is always the better option, so that incentive is really bogus. In the second case, if the product cannot earn its keep without monopoly, then why limit the patent term at all, or award the monopoly only to the inventor, and require novelty and non-obviousness? It is the third incentive that seems justifiable. But why not require that particular fact scenario to be in place rather than use novelty, etc. as the benchmarks for award of patent? Why not lmit the patent term to the time required to recoup the R&D? Why not extend the term if needed to ensure recoup of the R&D? Why not just replace the patent system with a grant system or a prize/bounty system? For example, presume the company does the R&D, brings the product to market, has the opportunity to see how profitable the product is in the presence of free market competition, and applies for a tax break in the amount of the loss. Does that not make more sense?
Did any US patents issue listing a slave as an inventor?
Interesting book. Automated inventing seems like a natural progression of software technology. I was recently reading about a company called Patent Express (www.patentexpress.com) that is automating the preparation and filing of patent applications. I think patent attorneys and people in our profession will gradually move to doing higher end tasks as well like client counseling, as the basic prep/pros work gets automated also.
Sooner or later, probably sooner, someone will file a patent with a program listed as the inventor. “The threshold question in determining inventorship is who conceived the invention.” If a program did the “conception”, it has to be the listed inventor. MPEP 2137.01.
Some of the machine learning programs are getting good at things like analog circuit design. Since they can conduct simulated experiments (using circuit simulators like SPICE)
This could be significant if a program makes a breakthrough in something like power supply design. That kind of design requires juggling more factors than a human can easily do, and most designs are suboptimal. Programs can potentially be better at this than humans.
6–
Your preliminary conclusion, that if the computer did not invent, then no invention took place, is unsupportable.
Since the final conclusion (that the computer invented) of your tautological argument rests on this unsupportable conclusion, it can be easily dismissed.
How can the question of inventorship be moot? I agree with you, though, that if there has in fact been any invention, certainly the operator of the machine is an inventor.
6: “The computer did invent, just because it brute force calculated the invention does not make it less of an invention.”
Hmmm. Debatable.
“The computer did not invent. It has no volition. It cannot conceive. It does not cognate. It is not autonomous. It exists because of, and acts according to the direction of, a conscious mind.”
If this were true then no invention took place. Since obviously invention did take place then this is false. The computer did invent, just because it brute force calculated the invention does not make it less of an invention.
The whole inventorship question is moot anyway. The person operating the machine can claim to be the inventor no matter what, nobody can gainsay him. He didn’t “derive the work from another” he merely made an observation in a natural habitat (his lab) that led him to conceive of the idea, and reduce it to practice first (i.e. he invented it). And this is true no matter if the computer handed him the idea or not. The dolphin handed the creator of the lunocet the idea of the fin needed to propel us at great velocity through the water. All he had to do was do some routine experimentation (no small task and it took awhile) as to using hinges etc. But nobody can contest that he invented the device, and that the prior art fins are woefully inadequate (due to the lack of flexibility iirc).
And yes MM you were right, he was asking a different question, I should have read more carefully.
Nice comment, Mortal. Going back to Latin roots, and “Invenire = to discover” maybe the human inventor is the intelligence operative who debriefs the computer, after it has run through its program, and picks out of the stream of data that the computer is debouching a selection deserving of the name “invention”? As you say, the “owner” is the one with the power to control access to what the computer is “setting forth”.
I guess the over-riding importance of who *is* the inventor is peculiar to US patent law. The real question must surely be: who is entitled to ownership of the rights?
Let’s assume that any novel, non-obvious invention that would be patentable if invented entirely by a human being ought to be patentable regardless of how it came into existence.
Suppose you developed a program that autonomously generated miscellaneous “inventions” without any human intervention once it had started running. Is there any reason to question that the devisers of the program ought to be the owners of the fruits of its execution?
Suppose you developed a program that generated “inventions” on the basis of information fed to it by human beings. Is there any reason to question that the devisers of the program together with the individuals who decided on the input ought to be the owners?
I’m sure that more highly nuanced scenarios could be posited, but I don’t see any great difficulty in applying the idea that whoever enabled the machine-conception of the invention should own the rights.
I also don’t see any particular problem with identifying the “humans-in-the-loop” as being the “authors”, to satisfy the US constitution.
clarification–
programmer+program=grad student
Possibly some unique capability in the grad student, combined with the appropriate technical ability, is the functional equivalent of the programmer and his/her program.
Anecdotally, there are many situations where grad students should be listed as inventors, but they are not, for various reasons. There are also many situations where the PI gives the grad student some vague criteria or initial conditions, then that student proceeds to do all the work and arrives at a solution, or type of solution, that was never actually contemplated by the PI.
Maybe that solution is patentable, or maybe it just forms the basis for more investigation that ultimately results in something patentable.
I have seen this happen, and then the fighting begins over who is named as an inventor.
Assuming the actions of the assistants mentioned are purely automatic, of course.
Luke–
Why could there not be multiple inventors?
At a minimum, I would suggest that both the programmer and the person who selected the particular data input would be co-inventors.
The programmer knew what the program would do in the general case, the person who selected the particular data input defined the parameters of the particular case.
Neither one acting alone would have achieved the particular result, but both acting together did.
The computer did not invent. It has no volition. It cannot conceive. It does not cognate. It is not autonomous. It exists because of, and acts according to the direction of, a conscious mind.
In a sense, it could be analogous to “the guys in the shop”, or the grad student, who actually executes the will of a researcher in achieving a particular solution.
I pretty well disagree with all of the above.
In my opinion, the main problem is likely to be one of identity of the “inventor”.
E.g. TJ wrote:
“I haven’t read the book and I am going to order it based on this post, but my initial reaction is, isn’t the bigger problem obviousness rather than subject-matter or utility. So I write a standard program that crunches the solution to a long-standing problem. The program is obvious; the solution it reaches is almost by definition obvious–if the person of ordinary skill is more than an automaton, then a fortiori automaton inventions are obvious.”
Where did your “standard program” come from? Probably the program will *not* be standard – the program will be the inventive part -or at least, for the initial use, anyway.
Being a chemist, I look at this from a chemical perspective.
In the current decade, an entirely new reaction was for the first time discovered “by” a computer.
The people involved collected a bunch of data and fed them into various ab initio calculations with various starting states. The computer churned through it all, and out came proposed reaction that nobody had so much as suspected could work – but in following up the suggestion, it was demonstrated to work just fine.
Clearly something – an entirely new class of reaction – has been invented here (which also will mean products produced therefrom)- but *who* invented it? The people who gathered the data that was fed into the program? Nah. The people who input the data? No. The people who read the output? No. The people who checked the results o confirm the prediction was correct? No. About the only candidate left is the people who “wrote the program”. But writing the program itself may well not be an inventive activity, as TJ posits. (There may be an analogy with copyright here. You seemingly can have “originality” in a computer-created work of art, even if all you did was enter the default starting conditions in drawing your Mandelbrot set…)
So what are we left with?
Unlike TJ, I don’t see how something – a entirely new class of reactions – that *no* person of *any* skill in the art could have contemplated, let alone one of ordinary skill, can be “obvious”. To me that is the epitome on NON-obviousness.
And yet he says that “the solution [an obvious program] reaches is almost by definition obvious”.
Convince me.
I think not. I think that, if you want to say such things shouldn’t be patentable, identity of inventorship is going to be a more fruitful bar than “obviousness” (or enablement, etc).
Cheers, Luke
“How about if the inventive product can be easily reverse engineered and manufactured by a copyist, such as the Chinese, and sold cheaper without the inventor’s R&D and advertising costs?”
A copyist such as the Chinese? LOL.
You know what the Chinese are copying now? The US business model of generating massive portfolios of crap patents. Isn’t that awesome?
But seriously, let’s take an item like candy. People still develop new candy all the time but it’s obvious stuff so you can’t get a patent. Can the Chinese make a cheap imitation? Probably. What if the imitation is identical or better and the Chinese do a better job of advertising?
Tough cookies.
Mooney said “As I’ve pointed out a billion times, however, that is only true where the cost of developing the new product is so high relative to the cost of manufacturing and marketing the product that a patent is necessary for anyone to profit from the sale of the product”
How about if the inventive product can be easily reverse engineered and manufactured by a copyist, such as the Chinese, and sold cheaper without the inventor’s R&D and advertising costs?
There is at least one patent (US 6,847,851) for a PID control circuit which was admitted by the listed human inventor to be developed by a software technique called genetic programming. The same technique has been applied to create patentable designs for antennas and a variety of electronic circuit hardfware with the only substantial input from the human inventor being the statement of the problem or constraints. See link to genetic-programming.org
and link to halfbakery.com
“Computers could care less.”
Couldn’t care less?
6, I think Captain Obvious is asking a more interesting (and rhetorical) question, i.e., if you took away patent protection for software, would a computer that does nothing but create novel software suddenly stop performing.
After all, the oft-repeated mantra around here is that “without patent protection, there is no incentive to innovate.” As I’ve pointed out a billion times, however, that is only true where the cost of developing the new product is so high relative to the cost of manufacturing and marketing the product that a patent is necessary for anyone to profit from the sale of the product.
Computers could care less.
“Do computer programs need patents as an incentive to invent?”
Yes, nobody would ever write programs if not for patent protection. You can see this because during the time period prior to it being adjudged patentable there were no programs written. I think. But then, one day the SC decided to open the door a crack and let people, who wanted to write programs but had no incentive to, write programs and thus Windows Vista was born. That’s the timeline in a nutshell.
Do computer programs need patents as an incentive to invent?
On a different note – Volokh is hosting a guest-blog this week by professor Mossoff regarding patent rights and sewing machines. The content appears fairly pro-patent and should interest several readers here.
link to volokh.com
I think the applicability of the utility requirement to subject-matter eligibility is an interesting topic. In fact, I wrote a law review Note that dedicated several pages to the subject a couple of years ago, 93 Iowa Law Review 1889 (2008), which is available here link to law.uiowa.edu
Don’t know if the book discusses, but there is a methodology out there with cult-like following called TRIZ. (link to en.wikipedia.org). A more mainstream company out of the same school is Invention Machine (http://www.invention-machine.com). Never quite got exactly what these guys did, or why they didn’t get more press.
He also left out enablement 🙁
“we need to strictly apply patent law’s utility requirement and its distinction between abstract ideas and practical applications [to] protect against the worst kind of abuses”
LOL. He left out written description. I wonder why?
I haven’t read the book and I am going to order it based on this post, but my initial reaction is, isn’t the bigger problem obviousness rather than subject-matter or utility. So I write a standard program that crunches the solution to a long-standing problem. The program is obvious; the solution it reaches is almost by definition obvious–if the person of ordinary skill is more than an automaton, then a fortiori automaton inventions are obvious.
Of course, this simply reveals in an extreme way the longstanding tension between IP law’s worship of creativity as against “sweat of the brow.” The input that is often lacking is not creative thought, but rather lots of money to buy computing power. At least, lots of computing power can substitute for creative thought. And patent law has often implicitly assumed a PHOSITA with almost infinite resources but no creative thought.
I have not read the book, I am just responding to the content of the post:
“…many products and processes used today were created through automated software.”
THROUGH, not BY.
More accurately, were created USING automated software as one of a possible plurality of tools.
What is the big deal? Why cannot hardware and software, and any integration between the two, be treated simply as a tool, like any other, that assists a person in the execution of his or her will?
I guess one would have to read the book to find out.
Comments are closed.